Western Michigan University

Covert Timing Channels Omar Darwish Instructor: Professor Elise de Doncker

Channels In Networks  Overt channel : ◦ Communication path designed to transmit the information in an authorized way.

 Covert channel: ◦ Information is transmitted in unauthorized way.

Covert Channels  Covert channel is an approach for leaking sensitive information across the network, by violating security policies.

 The covert channels are of two types: ◦ Covert storage channels.

 Depends in memory.

 Ex: Data is extra bytes added to ICMP error packets (May tell destination environment such as OS ) ◦   Covert timing channels.

Memoryless . For example transmit secret messages encoded by binary depends on interval packets delay

Covert Timing Channels

Covert Timing Channels  Channels that convey information through the arrival pattern of packets.

 Contents of the packets are not used.

Packet1 0 Packet2 1 Packet3 0 Packet4

Application of covert timing channels  Military application ◦ Passing secret messages.

◦ Covert channels detection applied by   USA government .

USA military bodies :National security agency, US Air Force, National Computer Security Center,  ◦ Intrusion attacks Viruses, hacking attacks

Covert Channels between Finite State machines   Suppose there are two finite state machines A and B that are connected via communication channel, such as the Internet.

A and B can exchange information freely without loss.

Cont…

 A third finite state machine C is an eavesdropper that knows all information about the finite state machines A and B , their transition functions, states, output functions, etc. In cyberspace.

 Finite state machines A and B can be considered as two online applications.

Cont…

   C can be an eavesdropper who can capture all information exchanged between A and B.

It is assumed that C knows alphabets, states, and normal transition functions of both machines A and B.

With this setting, there are several areas that A and B can take advantage of to establish covert channels so as to evade C.

A finite state machine (FMS)

 is a quintuple where If bandwidth of the covert channel is high, It means that it easy to detect.

Cont…

COVERT TIMING CHANNEL GENERATION TECHNIQUES AND TOOLS

Generating Covert Timing Channels State2 P :0.7-0.9

10-20 State3 P: 0.9-1.0

20-30 Int:erarrival times 2,3,5, 15 ,2, 25, 5,7, 19 010 State1 P: 0-0.7

0-10

Limitations of end-to-end encryption in secure computer networks by Padlipsky 1976

 The first paper introduces generating covert timing channel in 1976.

 Sender either send a message or keep silent during particular time.

IP Covert Timing Channels:

Design Cabuk 2004

Continue…

 Additional parity bits appended to the data.

◦ Redundancy for error correction due.

 Additional bits may be added for purposes of maintaining synchronization between sender and receiver  Finally, the data may be encrypted in order to add a further layer of privacy

Reducing Fuzzy Covert Timing Channels by Hu 1991  Categorize Timing Channels into :  Software Timing Channels.

◦ Processes communicate by modulating the amount of CPU time they use.

Hardware Timing Channel.

 Depends on traffic and bandwidth.

Ex: To send 1 generates series of instruction with bus traffic (such as referencing memory ).

To send 0 generates series of instruction with no bus traffic.

Cont…

 Covert timing channels has an accurate clock.

 The interval time representation depends on the current interval times.

COVERT TIMING CHANNEL DETECTION TECHNIQUES AND TOOLS

Detecting covert timing channels based on regularity by Cabuk et al. 2007  Standard deviation based: ◦ Low standard deviation  High regularity& High possibility of covert timing channel.

Fundamental regularity

Entropy

  Entropy is strongly tied with randomness factor.

The lowest the entropy is, the lowest of randomness in the stream

Examples of entropy

Ex: set of Data : 3,2,5,6,1 (Random) Entropy = -(1*Log (0.2) ) =0.6

Ex: set of Data : 3,2,3,6,3 (less Random) Entropy = -(0.6*Log(0.6)+0.4(log0.2)) =0.4

Entropy test score by Steven and Wang 2011  Entropy test score by Steven and Wang Low entropy  Low randomization & High possibility of covert timing channel.

References

1)Padlipsky,Limitations of end-to-end encryption in secure computer networks.1976

2) Girling , Covert channels in LAN’s, 1987 3) Hu,Reducing Fuzzy Covert Timing Channels, 1991 4) Cabuk, IP Covert Timing Channels: Design, 2004 5) Cabuk, Detecting covert timing channels based on regularity, 2007 6) Steven Gianvecchio, Model-Based Covert Timing Channels, 2008 7) Wu, Detecting IP Covert Timing Channels by Correlating Packet Timing with Memory Content, 2008 8) Liu, Hide and Seek in Time Robust Covert Timing Channels, 2009 9) Steven and Wang, Entropy test score 2011.

10) Wu, Identification and Evaluation of Sharing Memory Covert Timing Channel in Xen Virtual Machines, 2011 11) Yetiana, Constructing The On/Off Covert Channel On Xen, 2012 12) Sangamdace,Automatic Detection Of Illegal Transmission In A Network, 2012 13) Jonathan , Using Covert Timing Channels for Attack Detection in MANETs,2012 14) Daryl Johnson, Behavior-Based Covert Channel in Cyberspace, 2009.